From xemacs-m  Wed May 21 21:52:31 1997
Received: from mercury.Sun.COM (mercury.Sun.COM [192.9.25.1])
	by xemacs.org (8.8.5/8.8.5) with SMTP id VAA29293
	for <xemacs-beta@xemacs.org>; Wed, 21 May 1997 21:52:31 -0500 (CDT)
Received: from Eng.Sun.COM ([129.146.1.25]) by mercury.Sun.COM (SMI-8.6/mail.byaddr) with SMTP id UAA11360 for <xemacs-beta@xemacs.org>; Wed, 21 May 1997 20:07:20 -0700
Received: from kindra.eng.sun.com by Eng.Sun.COM (SMI-8.6/SMI-5.3)
	id TAA07194; Wed, 21 May 1997 19:51:48 -0700
Received: from xemacs.eng.sun.com by kindra.eng.sun.com (SMI-8.6/SMI-SVR4)
	id TAA14486; Wed, 21 May 1997 19:51:48 -0700
Received: by xemacs.eng.sun.com (SMI-8.6/SMI-SVR4)
	id TAA16746; Wed, 21 May 1997 19:51:48 -0700
Date: Wed, 21 May 1997 19:51:48 -0700
Message-Id: <199705220251.TAA16746@xemacs.eng.sun.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
From: Martin Buchholz <mrb@Eng.Sun.COM>
To: georgn@canada.Sun.COM
Cc: XEmacs Developers <xemacs-beta@xemacs.org>
Subject: Re: b1 success
In-Reply-To: <199705211433.KAA17537@verve.canada.sun.com>
References: <199705191905.PAA19737@verve.canada.sun.com>
	<199705200034.RAA12761@xemacs.eng.sun.com>
	<199705211433.KAA17537@verve.canada.sun.com>
X-Mailer: VM 6.31 under 20.2 XEmacs Lucid
Reply-To: Martin Buchholz <mrb@Eng.Sun.COM>

>>>>> "Georg" == Georg Nikodym <georgn@canada> writes:

>>>>> "MB" == Martin Buchholz <mrb@Eng> writes:
MB> Fixes appreciated.  However, note that gnudoit uses gnuslib.o
MB> which uses Xauth, which pulls in X, so that you're going to get a
MB> lot of the above libraries anyways.  Some of the libs come from
MB> LIBS_SYSTEM and friends in the s&m files, from which we haven't
MB> (yet) eliminated them.  Portably figuring out which libs we can
MB> leave out may not be the best way to improve XEmacs.

Georg> I've spent the last year of my life working in something we call the
Georg> Solaris ABI Program
Georg> (http://www.sun.com/developers/solbrand/app-cert.html, internal sun
Georg> people can look at http://abi.eng/) and we have developed tools to aid
Georg> in this kind of thing.  I won't bore you with arguments why I disagree
Georg> with your last sentence but I will leave you with the following
Georg> datapoint as food for thought:

It certainly would be better, in some sense, to split up the current
list of libs into more groups.  Currently there are only 2 groups: `X'
and `other'.

Georg> The gnudoit binary that I built from the last beta has 3714
Georg> dependencies on system supplied interfaces.  Only 37 of them are
Georg> direct.

Shouldn't the linker take care of making the overhead be link-time
only?  If I link with a library and use none of its symbols, should
the resulting binary even have a reference to that useless library?

Martin

